[PATCH] PR/16882: overload resolution of vector types

Paolo Bonzini bonzini@gnu.org
Thu Aug 5 15:34:00 GMT 2004


Paolo Bonzini wrote:

>> I'm of the opinion that this should be handled the same way we do
>> for regular arrays.  In particular the TYPE_MAIN_VARIANT should be the
>> array of the TYPE_MAIN_VARIANT of the element-type.
> 
> But if we do so, we print "const const __vector__ signed short" in the 
> case of vector types.

No, sorry, that's precisely what the patch does:

   tree innertype_main_variant = TYPE_MAIN_VARIANT (innertype);
   TYPE_MAIN_VARIANT (t)
     = type_hash_canon (hash, make_vector_type (innertype_main_variant,
                                                nunits, mode));

Paolo



More information about the Gcc-patches mailing list